From a theoretical standpoint, CBA is a tool for comparing options, particularly in the public sector. Public sector decisions are thought to be more complex because both policy and financial impacts must be considered. On the other hand, private enterprises theoretically need to be concerned only with the accountant's 'bottom line'. In reality, though, both private and public sector decisions are better when they consider all aspects of a given alternative, whether those aspects have a line in the balance sheet or not.
What we face is the difference between theoretical eloquence (which has its' weaknesses anyway) and real-world practicality. The goal is to find a practical way to accomplish the task at hand -- that is, performing the CBA -- while not opposing the basic theory of cost-benefit analysis. Note that this is not the same as embracing all the eloquence of theory. With this caveat, let us go on to the practical realm of CBA, pausing first to cover some past work in this area, and briefly, some basic cost-benefit theory.
There have been, though, very few papers written that offer a pragmatic outline of how to undertake a cost-benefit analysis for a geographic information system. Smith and Tomlinson (1992) present a method similar to the one offered here. Indeed, this paper marks an expansion of the method outlined in that work.
The benefit:cost ratio is just that -- a fraction representing the return (benefit) on the investment (cost). Usually the denominator -- the cost side -- is reduced to one (1). For instance, a $10,000 investment that returns $16,000 would have benefit:cost ratio of 16,000:10,000, or more commonly, 1.6:1. Any undertaking that offers a return greater than 1:1 satisfies one major requirement of economic justification: break-even -- that is, you won't lose money on the deal. However, there is another requirement, which is related to the economic concept of opportunity cost. Simply said, when deciding between two (or more) options it is not economically justified to select the option with the lower benefit:cost ratio. That is, if you have two options, one returning 1.6:1 and the other returning 2.1:1, the 1.6:1 option is not economically justified. The foregone opportunity (the one returning 2.1:1) is worth more. (Still, it may end up that the 1.6:1 is the preferred choice, because of other institutional, intangible, or non-economic considerations.)
The last issue is the future, and the impact it has on the numbers used in the CBA. Benefits in the future are valued less than benefits now. A simple example: compare getting $100 now or $100 next year -- the $100 now is preferred. The future benefits (and costs) must be discounted. Experience shows that a discount rate of 7 percent has proven reasonable (Nas (1996) validates this). This means that the evaluation is not $100 now versus $100 next year; instead it is a comparison of $100 versus $93 now. That is, the net present value (NPV) of $100 in one year is $93.
Thus, it is important to look at costs and benefits over time -- but not too long. There are too many uncertainties if the planning horizon is extended too far into the future. Hardware and software advances are expected to occur, but the impacts of those advances on the costs and benefits are just too uncertain. Experience shows that a six-year planning horizon is reasonable.
Over this six-year planning horizon there is a very typical pattern of costs and benefits. In the first years costs are high, and benefits are low. In the latter years benefits often greatly surpass costs. Thus, it is necessary to look at the cumulative effect of the costs and benefits over time. This is easily done once the cost and benefit numbers for each year are known. The cumulative numbers for the second year are the sum of the numbers for the first and second years, and so on until the sixth year, in which the numbers reflect the total costs and benefits for all six years.
In one recent project the benefit:cost ratio for the first year was 0.5:1, improving to 6.4:1 in the sixth year. Over the six years, this resulted in a cumulative benefit:cost ratio of 2.8:1.
The last manipulation of the annual benefits and costs involves discounting them over the six-year planning horizon. Because costs tend to be borne early and benefits late, and the value of future funds is less than that of current funds, the net result of discounting tends to be a downward force on the benefit:cost ratio. In the same project referred to above, the cumulative, discounted (at 7 percent) benefit:cost ratio for the six-year planning horizon was 2.6:1.
The project used here as an example shows fairly typical results. Experience has shown that a well-planned, carefully implemented GIS can achieve a long-term, fully discounted, cumulative return about 2.5:1 over the foreseeable planning horizon. This means that every dollar invested in hardware and software, data automation, system start up, application development, and the ongoing costs of operations can reap about $2.50 in benefits. Note that the 2.5:1 ratio is historic, and because every case is individual, no one can be assured of returns equal to this. In recent cases there have been returns ranging from less than 1:1 to 6.6:1. Individual projects have returned from 0:1 to nearly 20:1.
The planning effort should begin with a cataloguing of potential outputs. These outputs should be briefly described, and their potential benefits scoped out. This is an important step, which helps focus attention on those areas where GIS can offer the greatest impact. It can also avoid the problem of spending days defining and performing the CBA on an output that can ultimately generate only a trivial amount per year in benefits.
A second task to be done before individual outputs are defined is to catalogue the prices to be associated with costs and benefits. The most important thing is to define a salary scale for clerical, technical, professional, and administrative personnel. This is particularly important when working with multiple departments or agencies, because it keeps each unit's costs and savings on a comparable level.
When the client has a system in place or a specific system in mind cost determination is straightforward. In most strategic planning this is not the case, and the costs are determined by getting estimates from the major vendors. A variety of options should be considered. For instance, hardware options should include workstations and PCs, and include network considerations. The conservative approach is to use the highest estimate unless it is significantly beyond the scale of the majority of vendors. When considering staffing issues it is important to include employee overheads (on-costs) in addition to base salary. This will more accurately represent the cost of staffing.
In a large, government-wide project undertaken by Tomlinson Associates, benefits were grouped into four categories: direct, agency, government, and external. Direct benefits were those accruing as a result of using GIS rather than manual means for storing and producing information. Agency benefits reflected the increased productivity and improved quality of work credited to the better and more timely information produced by the GIS. Government benefits were similar to agency benefits, but considered how better information generated by one agency benefitted other agencies. Lastly, external benefits focussed on both qualitative and quantitative benefits that accrue outside the structure of government. A valid equivalent for the private sector would be direct, departmental, company-wide, and external.
Dickinson, Holly J. and Hugh W. Calkins, 1988. "The economic evaluation of implementing a GIS" in The International Journal of Geographical Information Systems, vol. 2, no. 4, pp. 307- 327.
Epstein, Earl F. and Thomas D. Duchesneau, 1984. The Use and Value of a Geodetic Reference System. Rockville, Maryland: National Oceanic and Atmospheric Administration, National Geodetic Information Center.
Nas, Tevfik F., 1996. Cost-benefit analysis: Theory and application. Thousand Oaks, California: Sage Publications, Inc. 220p.
Smith, Douglas A. and Roger F. Tomlinson, 1992. "Assessing costs and benefits of geographical information systems: methodological and implementation issues" in The International Journal of Geographical Information Systems, vol. 6, no. 3, pp. 247-256.
Wilcox, Darlene L., 1990. "Concerning 'The economic evaluation of implementing a GIS'" in The International Journal of Geographical Information Systems, vol. 4, no. 2, pp. 203- 210.
Yourdon, Edward, 1989. Modern Structured Analysis. Englewood Cliffs, New Jersey: Prentice-Hall, Inc. 672p.
Please contact the author for further information on the Tomlinson Associates projects referenced in this paper.